The data used in this paper have been obtained from Roberts, M. J. (1996) (Colombia, 1977-85: Producer Turnover, Margins, and Trade Exposure, Oxford University Press) with explicit permission from the authors. (allyears.dta)

Software used: MATLAB, R2015b, (8.6.0.267246) for .m and .mat files and Stata 12 for .do and .dta files
Operating system used: Ubuntu, 16.04 LTS
Expected computation time for the results below: <5 seconds

For the empirical results run the following codes:

- To replicate column 1 in Tables 2 and 3  load the data into Stata and then run the code Table_2_and_3_col1.do

-To replicate column 1 of Table 3 load the data into Stata and then run the code baseline.do

-To replicate column 2 of Table 3 load the file simulateddata.dta into Stata which contains the data from the baseline model simulation and run file simulatedregs.do

-To replicate the data statistics in Figure 2,  load the data into Stata and then run code Figure_2_data.do 

-To reproduce Figure 3, Panel A, load the data into Stata and then run code baseline.do to obtain age dummy coefficients for the actual data. Then load the file simulateddata.dta into Stata which contains the data from the baseline model simulation and run file simulatedregs.do to obtain age dummy coefficients for the simulated data. Then use Matlab file ageeffectsdummiessimdata.m produce the Figure 3, Panel A.

-To reproduce Figure 3, Panel B, load the data into Stata and then run stdsales.do to obtain the standard deviation of sales growth rates by age for the actual data. Then load the file simulateddata.dta into Stata which contains the data from the baseline model simulation and run file stdsalessim.do to obtain the standard deviation of sales growth rates by age for the simulated data.  Then using Matlab file  changeinstdevofsalesbyage.m to compute the change in the standard deviation and produce the Figure 3, Panel B.

-To obtain the mean standard deviations of growth rates reported in footnote 18, load the data into Stata and then run stdsales.do for the actual data and also load the file simulateddata.dta and run file stdsalessim.do for the simulated data.

-To replicate the regression of Table 4 of Appendix D, load the data into Stata and run robustness.do

-To obtain the correlation reported in footnote 23 of Appendix D, load the data into Stata and run the file lumpiness.do. The results are then imported to Matlab (file lumpiness.mat) and then run lumpinessbyind.m to obtain the reported correlation.


For codes used to solve the model, simulate and calibrate results, see folder ‘ComputationalCodes’ for further details.